How social identity and social diversity affect judging

Abstract Judges like to claim that they are impartial decision-makers fully capable of suppressing their personal proclivities, as the rule of law requires. But a century’s worth of studies undermines that view. Going under the name ‘judicial behaviour’, this vast literature shows that many extraneous (non-legal) factors affect the choices judges make. This article focuses on one strand of that literature – the effect of personal characteristics on judging, with emphasis on social identity and social diversity. We show that the literature is bifurcated: studies focusing on the social identity of individual judges (such as their gender, race, and nationality) generate findings consistent with in-group bias, whereas research on the social diversity of judges sitting in panels suggests that benefits can accrue from socially diverse courts. What the two sets of studies have in common, though, is just as important: both could make profound academic and policy contributions but require far more development if they are to realize their potential. We offer proposals for forward movement.


Introduction
When commentators tout the value of the rule of law, they often emphasize the importance of 'impartiality' in judicial decision-makingthat judges 'shall decide matters before them : : : on the basis of facts and in accordance with the law' 1 without regard to the identity of the parties.
For centuries this vision of impartiality has not only come to justify the rule of law and lend legitimacy to courts; it also lies at the core of most judges' self-image. Judges like to claim that they are dispassionate decision-makers fully capable of suppressing their 'personal proclivities' 2 And yet a century's worth of studies undermines this claim. 3 Going under the name 'judicial behaviour', this vast literature shows that many extraneous (non-legal) factors affect the choices judges make. 4 This article considers one strand of that literature, variously described as the analysis of the judges' social backgrounds, 5 personal attributes, 6 biographies, 7 or identities. 8 Whatever the descriptor, the basic idea is that personal characteristics affect judging.
The studies comprising this literature are both old and new. Early research tended to focus on career experience, asking, for example, whether former prosecutors-turned-judges are tougher on criminal defendants or whether judges who were corporate lawyers favour wealthy and powerful interests. 9 Partisan (political) identity too figured prominently in some of the original research, especially on US judges. Study after study found that judges appointed by, or affiliating with, the Democratic party, relative to Republican judges, were more supportive of workers, criminal defendants, and regulations on businessin other words, the Democratic judges were more left-leaning in their decisions. 10 Emphasis on career paths and partisanship continues today. 11 At the same time, as judiciaries throughout the world have grown (somewhat) more diverse in composition, 12  under analysis have expanded to include the judges' (and, sometimes, the litigants') social categories 13especially gender, 14 religion, 15 race, 16 ethnicity, 17 and nationality 18 Increasing attention to social identity and social diversity has contributed to the study of judging in more ways than one. First, because of world-wide interest in these twin topics, the field is more interdisciplinary than ever. In the not-so-distant past, the most prominent studies were authored by US political scientists studying US courts. 19 But these days work is just as likely to draw on economics, 20 law, 21 psychology, 22 and sociology, 23 as it is on political science; and research is increasingly global in scope, covering judges serving on domestic and international courts. 24 As diversity in approach, tools, and targets of inquiry has increased, so too has the depth of the studies and thus their capacity to enrich explanations of judging. 25 A second contribution traces to the studies themselves. Reading across the growing number of articles and books on social identity and social diversity in courts, it seems that their findings are convergingthough in distinct ways. On the one hand, research that characterizes individual judges on the basis of their social identity (gender, race, nationality, and so on) tends to generate results in line with in-group bias: the tendency of individuals to favour members of their own group over outsiders (such as, international court judges' favouritism toward their home government 26 ). On the other hand, research on the social diversity of collegial courts (those on which judges sit in panels or en banc) suggests that greater heterogeneity can produce benefits in the form of better decisions.
Characterized in this way, the two lines of research have quite different, even conflicting, implications for the legitimacy of courts in particular and the rule of law in general. At the level of the individual judge, differences in decision making based on social identity challenge an underlying justification for the rule of law: that it fosters impartiality. At the collective level of collegial courts, social diversity may enhance the quality of judicial decisions, thereby generating greater respect for judges and increasing the legitimacy of their courts.
These are the differences between studies on social identity and on social diversity. What the two have in common, though, may be even more important: Both have the potential to make important contributions to public policy.
Starting with work on the individual judge: if research can isolate features of judging that are most likely to exacerbate differences emerging from social identity, then scholars should be able to propose institutional reforms that will diminish their significance. An obvious example focuses on the connections among institutions (formal and informal rules) governing the appointment of judges, the people selected to serve and ultimately, the choices they, as judges, make. Understanding how the judges' personal characteristics affect their choices could inform institutional design by, say, identifying selection procedures that reduce the effect of partisan political identity on the judges' decisions or that lead to greater gender diversification. 27 Along somewhat different lines, research on social identity flagging particular biasessuch as the tendency of international court judges to favour their home government when it is a party to a dispute 28ought to force judges to confront the difficulties in suppressing 'personal proclivities ' and emotions. 29 Turning to collegial courts, research could determine whether greater social diversity, in fact, has salutary effects and what institutional practices are most likely to generate those effects. The basic task here would be to use empirical research to craft institutional procedures that foster positive collective benefits. 30 However conceivably important these policy implicationswhether for the social identity of individual judges or the social diversity of collegial courts -'conceivably' is the operative word: the implications remain more aspirational than realized. For even with the growth of research on the judges' personal characteristics, much more work is needed for the field to reach its full potential in terms of policy impact and we might add, academic consequence. For this reason, we propose opportunities for forward movement.
The discussion unfolds in two parts, reflecting the bifurcated approach (and implications) of existing studies of identity, on the one hand, and diversity, on the other. Section 2 focuses on identity, detailing research that characterizes individual judges as members of a particular group based on their gender, race, nationality, and so on. That literature, as we just suggested, generates findings consistent with in-group bias. Section 3, on diversity, considers studies of collegial courts. That work mostly reaches results consistent with the benefits that can accrue in socially diverse teams or groups, though again the research is quite nascent. Along the way, we point out gaps in  28 See the studies in supra note 18. 29 See, e.g., Wistrich, Rachlinski and Guthrie, supra note 2, at 909 ('Judges should be cognizant of their [reliance on the affect heuristic]. Most people fail to recognize its hidden influence. Awareness is not sufficient to ensure that judges keep emotional responses in check, but it is a necessary first step.'). the literature and offer suggestions for scholars hoping to contribute to these exciting areas of study.

Social identity and the judge
Prior to the 1980s, the systematic study of social identity was difficult because courts in many countries were quite homogeneous. Consider, for example, the judges' gender. As Figure 1 shows, 31 through the 1970s only about 12 per cent of the 155 countries represented in the data had selected a womantheir first womanto sit on the country's highest court; in the 1980s, that percentage doubled, such that at least one woman had served on about (12.312.9=) 25 per cent of the countries' courts. By the end of 2010s, only four countries had never appointed a woman to their highest court. This is not to say that all high courts are now composed of substantial percentages of female judges. On the contrary: the fraction varies considerably. For international courts, Grossman's (1999Grossman's ( -2015 data show that women occupied 47 per cent of the seats on the International Court of Justice but only 14 per cent on the European Court of Justice. 32 Valdini and Shortell's (2017) data for domestic high courts also shows considerable variation, from Latvia's 57.1 per cent female to under 10 per cent for 11 countries (including Italy, Panama, India, and the UK). 33 Nonetheless, even low percentages can mask growth in gender diversity over time. In Latin America, for example, the number of female justices 'increased dramatically : : : from 3% of all justices in the region's high courts in 1980 to 19% in 2010'. 34 Quantifying increases in other social categories (such as race, ethnicity, and religion) is harder because of varying cleavages on the relevant identities, as well as a lack of data on judge characteristics. 35 But a look at religion in Israel and the United States may suffice to make the point. As Figure 2 (left panel) shows, 36 in 2010 religious Jewish justices (relative to secular Jewish and non-Jewish justices) cast fewer than 25 per cent of the votes in the Israeli Supreme Court; a decade later, the percentage jumped to 41. Growth in religious diversity on the US Supreme Court is equally apparent (right panel). Between 2000 and 2020, the percentage of votes cast by Catholic justices (relative to all other religions) doubled, from about 33 to over 66.

The studies
Increasing social diversity among judgesnot just on gender and religion but also on race, nationality, and ethnicitynaturally enough generated research centred on whether the particular category 'mattered'. Do female judges 'speak in a different voice' than their male counterparts? 37 Are Black US judges inclined to treat Black defendants more leniently (and likewise for white judges/ defendants)? 38 Compared to less devout judges, do religious judges more frequently support claims of religious liberty? 39 The diversity turn also triggered new research programmes on courts  The 'different voice' language is from C. Gilligan, In a Different Voice: Psychological Theory and Women 's Development (1982). In the judging-gender literature, the expectation from Gilligan's work seems to be that female judges will bring a 'feminine perspective' to the benchone that 'encompasses all aspects of society, whether or not they affect men and women differently' and not only 'the political agenda associated with feminism'. S. Sherry, 'Civic Virtue and the Feminine Voice in Constitutional Adjudication', (1986) 72 Virginia Law Review 543. As discussed below, empirical studies do not seem to support the hypothesis of differences between male and female judges across all issues. that always have been socially diverse on some dimensionon the European Court of Human Rights, for example, nearly 50 different nationalities have been represented at one time or another. 40 But studies systematically exploring whether national identity (or ethno-national affiliation) affects the choices of judges serving on international and domestic courts have grown in number and sophistication in recent years. 41 Theoretically, projects attempting to answer questions about (possible) differences based on social categories are all over the map, offering an array of explanations. 42 The findings, though, are more uniform, with many fitting comfortably with one of the most central and bestdocumented manifestations of social group identity: in-group bias (or favouritism). More than five decades ago, social scientists noticed the tendency of individuals to favour members of their own group over outsiders; 43 and now there is a very substantial literature on the subject across law and the (social) sciences. 44 The result is an impressive body of evidence showing that people tend to be more helpful, more willing to allocate resources, and more supportive of policies advocated by members of their own group (and vice versa for the outgroup). Moreover, the greater the salience and relevance of the grouping to the task at hand, 'the stronger these intergroup divisions'. 45 Evidence consistent with in-group bias in courtsthe 'us-against-them' judging at the heart of the problematic relationship between impartiality and judicial decisionsis far and wide. 46 Starting with gender, contemporary studies have converged on results showing that female judges are more favourable toward plaintiffs (mostly women) in cases of gender-based discrimination and sexual harassment. This holds on courts as diverse as the US intermediate appellate courts, 47 the European Court of Human Rights, 48 and the Supreme Court of Canada. 49 Then there is evidence in the family law context of female judges tending to favour mothers over fathers, whereas male judges support fathers over mothers; 50 likewise, in criminal cases involving sex offences (again, 40 Data supplied by Eric Voeten at Georgetown University. 41 See the studies listed in supra note 18. 42 The accounts range from the representational (judges serve as representatives of their social group and so work to protect it in litigation of direct interest) to the informational (judges of different social groups possess unique and valuable information emanating from shared experiences) to the organizational (judges of all social groups in a society undergo identical professional training, obtain their jobs through the same procedures, and confront similar constraints once on the bench). See Boyd, Epstein and Martin, supra note 14. There are exceptions. For example, in a study of decisions in immigration cases, Gill, Kagan and Marouf, supra note 23, find effects seemingly at odds with in-group bias. As the authors explain, 'Masculinity theory predicts that men will be disadvantaged when appearing in front of an all-male panel, while chivalry theory predicts that women will benefit from an allmale panel. Our results provide support for both of these theories.' with mostly male defendants), female judges favour the state at greater rates than males. 51 Beyond these types of casesall salient to genderfew (or weaker) differences emerge between male and female judges. Running along similar lines are studies of the judges' (and litigants') race in the United States. To supply but a few examples: White judges are less likely than Black judges to hold for Black defendants claiming police misconduct; 52 Black federal appeals judges, relative to whites, tend to support plaintiffs of colour in voting rights cases 53 and Black claimants in employment discrimination litigation; 54 and white judges are less supportive of affirmative action (diversity) programs than Black judges. 55 Worthy of mention too is Sen's research showing a greater willingness on the part of (mostly white) US appellate judges to reverse the decisions of Black trial court judges. 56 Studies of religion and nationality also shore up various forms of us-against-them judging. In separate papers, Weinshall (Israeli justices), and Epstein and Posner (US justices) find a relationship between devoutness and support for religion: 57 Religious Jewish justices on the Israeli Court are far less likely to overrule decisions of the rabbinical court than secular Jews; and highly devout US justices (many of whom are Catholic) more frequently rule in favour of religious liberty than the less devout. 58 Koev's study of religious freedom cases in the European Court of Human Rights also points to a form of us-against-them judging: the judges are less likely to rule in favour of Muslim applicants. 59 Whether this result reflects bias against adherents of Islam is hard to know, as Koev acknowledges. Nonetheless, 'the strength of the empirical relationship, when paired with the salient and controversial role Muslims (especially Muslim migrants) play in Europe's contemporary political climate, suggests such a bias is plausible'. 60 In-group favouritism based on national identity is equally evident. Empirical studies give little reason to doubt 'that the nationality of the international adjudicator matters'. 61 To provide an (extreme) example, an Italian judge on the European Court of Human Rights dissented in 133 judgments concerning alleged Italian violations of the European Convention on Human Rights. 62 Even on domestic courts (ethno-)national affiliation can lead to in-group bias, as a study of the Constitutional Court of Bosnia-Herzegovina uncovered: 'the judges : : : divide predictably along ethno-national lines [and] these divisions cannot be reduced to a residual loyalty to their appointing political parties'. 63 51 Weinshall, supra note 39, for example, finds that female judges on the Israeli Supreme Court, 'were significantly "tougher on crime" when sex offenses were involved, ruling for the state in 80.2% of their decisions, compared to 67.1% of the decisions by men'.

The gaps
However promising this line of research, much more work must be done to develop a fuller picture of the relationship between social identity and judging. Four gaps come to mind.
First, although many results are consistent with in-group bias accounts, 'consistent' is worthy of emphasis. With very few exceptions, 64 the researchers did not set out directly to assess us-againstthem judging. They instead sought to answer questions that focus on difference (e.g., do female judges speak in a different voice?), and not on the judges' treatment of out-versus in-groups.
The exceptions, though, are instructive. Take Shayo and Zussman's famous study of small claims courts in Israel, where cases are randomly assigned to the judges. 65 The analysis unearths clear 'us-against-them' judging based on religion and ethnicity: Jewish judges systematically favour Jewish litigants, and Arab judges favour Arab litigants. Further, when the salience of these identities increases, say, in the aftermath of terrorist attacks, the bias strengthens.
Because of its careful attention to theory, design, and analysisnot to mention the researchers' substantive knowledgethis study provides a clear roadmap for future work on judicial in-group bias. 66 Indeed, to the extent that in-group bias is a feature of human cognition, Shayo and Zussman's findings should transport to other societies, though the relevant 'teams' may differ. In the US, for example, where the American public is highly polarized along partisan lines, one might expect judges to be biased in favour of co-partisans and against opposing partisans in, say, disputes implicating elections and voting.
Second and relatedly, assuming judges are no more immune to in-group bias than the rest of us, the mechanism remains unclear. The chief hurdle is that in-group bias could be a manifestation of judges rationally pursuing their self-interest, for example, voting for people like them to advance their future job prospects; or of judges 'thinking fast', 67 that is, relying on heuristics, emotions, intuitions, and the like to make quick decisions without much effort. In other words, looming large is the problem of behavioural (or observational) equivalencewhen different accounts generate the same prediction about the judges' behaviour.
By way of example, consider Posner and de Figueiredo's study on the International Court of Justice (ICJ). 68 After demonstrating that ICJ judges tend to vote in favour of their home country (a form of in-group bias), the authors offer an explanation grounded in (self-interested) rational choice theory: Economically, judges may be motivated by material incentives. Judges who defy the wills of their government by holding against it may be penalized. The government may refuse to support them for reappointment and also refuse to give them any other desirable government position after the expiration of their term. 69 But another mechanism is equally plausible: ICJ judges side with their own country not because they are rationally advancing an economic or any other interest but because of an emotional response. Posner and de Figueiredo recognize as much when they offer this alternative explanation for their finding: 64 Wistrich, Rachlinski and Guthrie, supra note 29; Epstein, Parker and Segal, supra note 44; Shayo and Zussman, supra note 15. 65 Shayo and Zussman, ibid. Psychologically, if judges identify with their countries, they may find it difficult to maintain impartiality. ICJ judges are not only nationals who would normally have strong emotional ties with their country; they also have spent their careers in national service as diplomats, legal advisors, administrators, and politicians. Even with the best intentions, they may have trouble seeing the dispute from the perspective of any country but that of their native land. 70 Differentiating between these two possible mechanisms is not easy because either could generate the observed results: in-group favouritism. With clever designs and data, however, we can and, more to the point, must solve the equivalence problem if we are to develop more comprehensive explanations of the role of social identity in judging.
A third gap in the social identity literature squarely implicates 'identity' or more accurately 'identities'. Virtually all existing data studies of judging 'separate out the effects of individual identity attributes'. 71 A Jewish female judge is decomposed into a Jewish effect and a female effect; a Black male British judge into a Black, a male, and a British effect, and on and on. Many scholars working in the diversity space, however, reject this approach arguing instead that 'the inseparability of attributes : : : should give us pause when interpreting data sorted by a single attribute '. 72 Recognizing that individual judges are, like all of us, bundles of identitiesidentities that intersect and overlap 73is crucial to advance work in the field. But the studies need not move in lockstep. One approach, following from research on social categories (with Shayo's theoretical work especially informative 74 ), is to consider more carefully when one identity over others is likely to be activated because it may be particularly salient to the dispute at hand. Another is explicitly to account for the intersectionality of identities in the research. 75 One of the rare examples is a study of US appellate judges that models the joint effects of gender and race. 76 A final possibility for future work entails a synthesis of research on judicial selection mechanisms and research on social identity. Existing studies on the effects of selection mechanisms have focused on the general question of whether different mechanisms produce differences in decision making. The studies have yet to drill down to investigate the substantive content of those differences. Future work on the institutional design of judicial selection should contemplate whether different methods of appointment (and retention) result in judges who are more or less prone to these forms of in-group bias.

Social diversity on collegial courts
Depending on your perspective, judicial in-group bias could be seen as offsetting years of 'exclusionary legal processes' 77 or it could be seen as violating a guiding principle of most courts: to treat all parties equally. Either way, a mix of social identities is essential for ensuring socially diverse courts ('teams of judges'), which, in turn, may be crucial for developing innovative, high-quality solutions to the kinds of complex problems that confront contemporary courts. In other words, a 'diversity bonus' 78 may follow for courts with a mix of social identities. 79 Why? Diversity theorists posit several mechanisms. Most well suited to courts, we believe, are approaches that emphasize shared information and the requirement of reaching an agreement on a collective decision. Scott E. Page, for example, argues that the key to building great teams lies in germane cognitive diversity, which amounts to 'differences in information, knowledge, representations, mental models, and heuristics' that team members bring to the tasks of 'problem-solving, predicting, and innovating'. 80 To the extent that social category diversitydifferences in race, gender, ethnicity, religion, nationalityfeeds into people's 'cognitive repertoire', socially diverse groups will perform better.
Another prominent scholar, Katherine W. Phillips, agrees with Page but more directly connects social diversity and 'smarter' decision making. As she puts it: The key to understanding the positive influence of diversity is the concept of informational diversity. When people are brought together to solve problems in groups, they bring different information, opinions and perspectives. This makes obvious sense when we talk about diversity of disciplinary backgrounds-think [of an] interdisciplinary team building a car. The same logic applies to social diversity. People who are different from one another in race, gender and other dimensions bring unique information and experiences to bear on the task at hand. A male and a female engineer might have perspectives as different from one another as an engineer and a physicist-and that is a good thing. 81 On this account, social diversity leads to better decisions as people bring different perspectives to bear on the problem at hand; in other words, the more diverse the inputs, the stronger the outputs. Benefits also accrue, Phillips maintains, 'from simply adding social diversity to a group' because 'people believe that differences of perspective might exist among them and that belief makes people change their behavior' by working harder and 'encouraging the consideration of alternatives before any interpersonal interaction takes place'. 82 The mechanisms identified by Page and Phillips are especially relevant to analyses of collegial courts. Diverse judges likely bring different experiences and perspectives to the factual and legal questions they confront, and the potential influences of those differences are numerous. They may affect the ways in which a collegial court identifies and weighs the controlling facts in a particular case. They may influence how the court assesses the implications of the relevant 'law' (broadly defined to include constitutional provisions, statutes, past judicial decisions, and the like) for the kinds of subsequent behaviour its decisions might induce. Or they might shape the court's collective assessment of what a fair or just outcome entails. 77 Escobar-Lemmon et al., supra note 27. 78 Page, supra note 71. See also Page's The Difference: How the Power of Diversity Creates Better Groups, Firms, Schools, and Societies (2007). Section 3.2 reviews empirical studies supporting this claimand how the lessons from those studies could be applied to courts.
These are just a few of the potential effects of diversity on the collective decision-making process of a collegial court. Whether or not they, in fact, enhance collective decisions in the beneficial ways anticipated by the 'diversity bonus' literature is an important subject for empirical research.

The studies
Although research directly putting ideas about the value of diversity in the courtroom to the test is scant, a handful of results could be seen as consistent with them. These results follow from 'The Collegial Court' studiesa line of inquiry that seeks to assess whether a case's outcome (or a judge's vote) would have been different had a single judge, and not a panel, decided the case. In other words, the key question is whether 'collegial ' (or 'panel' or 'peer') effects exist. 83 Often the focus is on the social diversity of the panel, especially its racial 84 or gender 85 composition. 86 Regardless of the specific social grouping under analysis, the studies tend to support Phillips's claims about the value of informational diversity: Overall, they find that 'minority' judges on a panel can affect the choices of their colleagues in pertinent areas of the law because those judges possess information, experience, or expertise that is valuable to their colleagues. Research on US appellate judges, for example, demonstrates that the presence of a Black judge on an otherwise white panel, leads the white judges to issue decisions more favourable to Black plaintiffs in areas of the law where race is prominent (such as, voting rights and affirmative action). 87 Work on gender finds similar collegial effects. Panels with one of more female judges on international criminal tribunals give substantially longer sentences to sexually violent offenders than all-male panels (about 35 months longer); 88 and males serving on US appellate panels are significantly more likely to hold in favour of plaintiffs in gender-based employment cases when a female serves on the panel. 89 Also sitting comfortably with the importance of diverse perspectives are reports that gender-and racially-diverse panels more frequently consider alternative views on the questions presented in cases. 90

Proposals for forward movement
Our review of the relevant studies is, admittedly, brief but for a good reason: Researchers have barely scratched the surface when it comes to using data to suss out diversity 'bonuses' on collegial courts. The basic question for our purposes is, what constitutes a 'better' judicial decision or a 'better' court? The list of possible benefits (forms of 'better') worthy of exploration is long but includes equity, legitimacy, quality, innovation, and creativity. 83 The first sense of 'better'equityaligns with justifications for the rule of law: Are the decisions of collegial courts more impartial than those of individual judges? Studies of equity effects on collegial courts answer this question in the affirmative, suggesting that socially diverse courts may mitigate in-group bias. Recall, for example, the study showing that in family law disputes male judges tend to favour fathers and female judges favour mothers. The same study, though, reports that 'these effects are dampened on panels that include judges of both genders'. 91 Likewise, a study of sentencing in the United States finds 'that as the proportion of Black judges increases, white and Black judges are less likely to render incarceration sentences in cases with Black defendants and white judges are more likely to render incarceration sentences in cases with white defendants '. 92 Both studies, in other words, imply that social diversity can redress inequities caused by in-group bias. But far more work is needed, especially on courts outside the United States.
A second, though related, sense of 'better' is grounded in questions of the legitimacy of courts and how social diversity may produce possible legitimacy benefits. To be sure, there are many claims to the effect that 'An all-male bench is no longer legitimate' 93 or that social-identity 'representativeness is a democratic value that can serve to justify the [courts'] exercise of authority'. 94 Unfortunately, though, empirical support for these claims is limitedthough not non-existent. In a notable experiment, researchers showed participants one of two articles: one reporting that Blacks comprise 23.2 per cent of the US federal bench and the other reporting that only 3.9 per cent of federal judges are Black. 95 Black respondents shown the '23.2 per cent' article were more likely to respond positively to questions like 'The courts can usually be trusted to make the right decision.' Another experiment, this one on gender, found that the presence of women on a committee (perhaps akin to a judicial panel) conferred greater 'institutional trust and acquiescence' to its decisions. 96 The same experiment also demonstrated, intriguingly, that women's presence had a legitimizing effect (especially on men) on decisions that 'go against women's interests', such as decisions that question policies on sexual harassment.
Beyond these, empirical studies of the effect of social diversity on legitimacy are few and far between. Not even research specifically aimed at identifying the correlates of trust or confidence in judiciaries explicitly considers the diversity of the courts under analysis. 97 With effort and creative thinking, this gap could be filled. Survey-experiments can be conducted (relatively cheaply and rapidly) using participants recruited via Amazon's Mechanical Turk (MTurk). Although not without its share of problems, used with care MTurk can produce reasonably high-quality data. 98 Building the judges' social identities into cross-national studies of 91 Stribopoulos and Yahya, supra note 11. 92 confidence, trust, and legitimacy is also possible, assuming substantive knowledge of the legal systems and of the relevant cleavages, along with careful data collection and analysis.
A third sense of 'better' refers to the ways in which collective decision making enhances the substantive quality of decisions. Questions of quality, as well as innovation and creativity directly implicate the benefits of diversity espoused by Phillips, Page, and other scholars 99and those benefits have been well documented both in the business world and in academia. Studies show that family-owned companies with at least 10 per cent female executives substantially outperform male-only companies by over 400 basis points per year; 100 that gender-and/or ethnically-diverse companies are significantly more likely to introduce innovations than firms dominated by one gender or ethnicity; 101 and that higher levels of racial diversity in firms are associated with more revenue, such that 'the mean revenues of organizations with low levels of racial diversity are roughly $51.9 million, compared with : : : $761.3 million for those with high levels of diversity'. 102 As for the academy: Freeman and Huang, among others, demonstrate that scientific papers authored by ethnically diverse teams are more consequential and impactful; 103 and that 'greater homophily is associated with publication in lower-impact journals and with fewer citations'. 104 Although research running along similar lines has been conducted on juries, 105 studies of courts are almost non-existent. 106 We should rectify that. Just as scholars have used citation counts to measure the quality of papers produced by research teams that are socially diverse versus those that are homogeneous, 107 we could do the same with court decisions produced by socially diverse panels versus those that are not. Indeed, scholars have long used citations to judges' opinions to assess the judges' quality. 108 So too research might consider whether decisions issued by diverse teams of judges are less likely to be reversed by higher courts. Finally, projects on the relationship between social diversity and judicial innovation also could be adapted from existing research, including work that traces the development of a 'new' rule or standard and tracks its diffusion (or lack thereof) across courts and societies. 109 4. From promise to reality: Diversity is the key Our primary goal in this essay was to offer what we think (hope!) are promising avenues for future work on social identity and social diversity. But, as we have signaled throughout, the studies will not write themselves; hard work and creativity are required to move in the proposed directions. That work is likely to come in many different forms, using data developed from in-depth interviews, structured surveys, experiments, court records, and so on. Moreover, researchers may well confront obstacles along the way, such as preserving the privacy of litigants and judges. 110 We are agnostic about the form of the research; data are data and methods are methods. 111 Which types get selected (should) depend on the researchers' questions and goals.
Where we feel far more strongly is in how scholars go about building collaborations. After all, an important implication of the research we have reviewed is that scholars should develop diverse research teams. Relevant substantive diversity, of course, is crucial. A survey researcher, a specialist in judicial behaviour, and an expert in law are more likely to generate higher quality work products than, say, a team of all survey researchers or of all specialists in judicial behaviour. But, as Phillips suggests, a male Norwegian survey researcher and a female Taiwanese survey researcher are likely to bring different perspectives to the table with similarly beneficial results.
Building social diverse teams may not be easy, but the lesson from decades' worth of research is that the benefits may well outweigh the costs. To once again quote Phillips, 'The pain associated with diversity can be thought of as the pain of exercise. The pain : : : produces the gain. In just the same way : : : we need diversity if we are to change, grow and innovate.' 112 This holds for all teams and all organizationscourts not excepted.